"Message Too Long" error on extremely short prompts (e.g., "1+1=?") & Severe Context Corruption

[Ashim_Credeor]

Describe the Bug

I am experiencing a critical issue where the chat refuses to process even the shortest prompts, claiming the message is too long. Additionally, when the AI does respond, the answers are incoherent or completely irrelevant to the context.

Steps to Reproduce

write 1+1

Operating System

Windows 10/11

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.1.36 (system setup)
VSCode Version: 1.105.1
Commit: 9cd7c8b6cebcbccc1242df211dee45a4b6fe15e0
Date: 2025-11-26T03:43:48.079Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Windows_NT x64 10.0.26200

For AI issues: which model did you use?

auto

Does this stop you from using Cursor

yes. i cant use cursor

[deanrie]

Hey, thanks for the report. This is definitely a bug - a simple “1+1” shouldn’t cause a “message too long” error.

To help investigate, please provide:

1. Request ID: Open the chat context menu (top right) → Copy Request ID, then paste it here
2. Console logs: Help → Toggle Developer Tools → Console tab - screenshot any errors when this happens
3. Context details:
   • How many files are attached or in context?
   • Do you have a .cursorrules file? If yes, how large is it?
   • MDC rules: Do you have specific rules that auto-attach to the chat?
   • How many files are currently open in your editor?
   • What’s the total size of your project?
4. MCP servers: Do you have any MCP servers configured? (Cursor Settings → MCP & Tools)
5. Empty project test: Create a new empty folder with no .cursorrules, open it in Cursor, and try “1+1” in a fresh chat - does it still happen?

Temporary workaround: Try creating a new chat with all files detached (clear context) to see if that works.

This matches a pattern where large context or rules can unexpectedly exceed token limits. The diagnostic info will help the team understand the cause.

[Ashim_Credeor]

1. request id: 595a9386-96d9-4da2-9c18-689695b6626e

2. console is not activate

3. no context, no files. /
   SOLID principles must be followed.

Identify the root cause of the problem. Avoid creating patchwork code.

Before removing any code, always check for side effects.

Identify connectivity issues first.

Summarize to the essential points.

No emojis.

Provide only factual information.

Avoid rhetoric.

Record and update project progress through external files.

Avoid making unconfirmed promises (e.g., “I can now open and run my project in Godot!”, “My project will now run properly”).

Always respond in Korean/
no mdc/
only one file in editor/
1200+ code page(java)

4. no mcp
5. sometimes work(15%)

[deanrie]

Thanks for providing the request ID! Based on your details, the large Java file (1200+ lines) combined with your .cursorrules is likely causing the token limit to be exceeded even with simple prompts.

Immediate solutions:

• Detach the large Java file: Close it in your editor
• Switch to a larger context model: Try Claude Sonnet 4.5 1m or Gemini 3 with max mode - they handle more context
• Test with extensions disabled: Run cursor --disable-extensions in terminal (based on a similar issue)

Recommended solution:

• Split your large Java file into smaller classes/modules - 1200+ lines in a single file is difficult for AI models to process efficiently. Breaking it into logical components will solve this issue permanently.

Additional diagnostics:

• Could you please share console logs? Help → Toggle Developer Tools → Console tab
• Try sending “1+1” again
• Screenshot any errors

Let me know if switching models or detaching the file helps.